Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Sample size for four-level logistic regression


From   Phil Schumm <pschumm@uchicago.edu>
To   <statalist@hsphsun2.harvard.edu>
Subject   Re: st: Sample size for four-level logistic regression
Date   Fri, 21 Jun 2013 12:48:09 -0500

On Jun 21, 2013, at 12:20 PM, Phil Schumm <pschumm@uchicago.edu> wrote:
> Alternatively, you could proceed as follows.  Generate data under the full model, and then fit two models: the simple model I described at the outset, and the full bells-and-whistles mixed model.  This will give you an estimate of the relative precision of the two estimates of the effect of the intervention (i.e., one from each model), which you can refine by doing additional simulations (but a much smaller number than estimating the power directly).  You can then use this estimate of relative precision to adjust your initial power estimate (based on the simple model).


FYI, I should have clarified the fact that in the simple model, you are estimating the marginal effect of the intervention, and this is likely smaller in magnitude than the effect you are estimating with the mixed model.  Thus, you need to take this into account when you compare the variances of the estimates from both models (i.e., the two estimates you compare have to be estimating the same thing).  One way to do this would be to use -margins- following the second model.


-- Phil


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index